[SOLVED] Best way to test RAM without adequate CPU cooling?

[OUTBURSTPAL]

Hey guys, I just got new RAM because my old RAM was not on the QVL of my mobo, and was casing major instability issues. (Random shut down, restart, BSOD etc. Another interesting one was when I had the PC asleep, sometimes when trying to boot back up the memory would become undetected and show error 55 requiring me to to unplug and replug in PSU. Anyway, I got my new RAM 2 days ago. The most I have been able to do to test is keep the PC idle on desktop due to me not being able to cool the CPU properly. I have the waterblock on the CPU, but my GPU is currently being fixed and I have hardline tubing so I can’t exactly just setup a temporary cooling solution. Anyway, I remember with my old RAM even just being idle on desktop would cause issues, I remember when first building, I experienced a random shut down after the first 20 minutes or so of the PC being on. The total amount of time I have had the new RAM essentially active probably adds up to around 24-36 hours (including sleep time) So basically that long with no issues. so with the new RAM I have had periods of about 30 minutes having it idle on desktop then shut down and let it cool down for an hour. Then repeat. Haven’t had any issues yet which is promising, but I was wondering if there was a better way to test it without having adequate cooling? I have also been leaving it on sleep a lot to see if I could replicate the “memory undetected” error when waking up. Seems okay so far, but I’m not sure whether the RAM has to be put under real stress for it to gradually worsen the symptoms. Anyway, the main reason I’m testing the new RAM is just because I want to be 10000% sure it was the old RAM casing the issues and not the motherboard. Although I am pretty confident it was the old RAM.

 

[OUTBURSTPAL]

So I was just thinking. Since technically the RAM is still in use when the PC is asleep, some particular issues that would happen while the PC is running would also happen when in sleep mode right? Like with my old RAM when it became undetected when waking up, was that probably due to something happening when in sleep mode? So I guess what I’m trying to ask is having it in sleep mode a good enough test?

 

[Rogue Leader]

The ram needs to be actually stress tested, running at idle or asleep is NOT a sufficient test for stability at all.

You cannot do this without proper CPU cooling, full stop. You WILL damage your processor.

You need to wait until you can get your system back together properly, or swap your water block out for a cheap air cooler.

 

[OUTBURSTPAL]

The ram needs to be actually stress tested, running at idle or asleep is NOT a sufficient test for stability at all.

You cannot do this without proper CPU cooling, full stop. You WILL damage your processor.

You need to wait until you can get your system back together properly, or swap your water block out for a cheap air cooler.

Guess I’m just gonna have to wait till my system is back together and hope for the best 🙂

 

[OUTBURSTPAL]

The ram needs to be actually stress tested, running at idle or asleep is NOT a sufficient test for stability at all.

You cannot do this without proper CPU cooling, full stop. You WILL damage your processor.

You need to wait until you can get your system back together properly, or swap your water block out for a cheap air cooler.

I do have a question though. How likely is it that the motherboard itself being defective would cause such generic RAM issues? For example, say the sticks don’t work properly in dual channel mode, but when disabled (put side by side in slots B1 and B2 they work perfectly normal. This was the case with my old Ram. Ram issue or mobo ?

 

[Rogue Leader]

I do have a question though. How likely is it that the motherboard itself being defective would cause such generic RAM issues? For example, say the sticks don’t work properly in dual channel mode, but when disabled (put side by side in slots B1 and B2 they work perfectly normal. This was the case with my old Ram. Ram issue or mobo ?

Unlikely its the motherboard.

 

[OUTBURSTPAL]

Unlikely its the motherboard.

Phew, thanks. Gives me a peace of mind since it’s a $1600 motherboard.

 

[Karadjgne]

More likely it's the memory controller in the cpu. If the cpu block is installed incorrectly, it can apply uneven pressure on the socket and especially on the right side, it'll cause ram errors galore, even to the point of boot refusal.

Also, depending on the ram itself, the bios may not be supplying enough voltage to the ram in dual channel mode, xmp is a generic setting and some ram has slightly substandard chips that require slightly more voltage than XMP will set. I've seen xmp set 1.35v, and the ram be unstable/non working, but manually raising to 1.37v allowed perfect stability.

And totally agree with the advice to wait. You Cannot diagnose full functionality of a component on a pc that isn't fully functional. That's like trying to guess how well a car will perform when it's missing a motor...

 

[OUTBURSTPAL]

More likely it's the memory controller in the cpu. If the cpu block is installed incorrectly, it can apply uneven pressure on the socket and especially on the right side, it'll cause ram errors galore, even to the point of boot refusal.

Also, depending on the ram itself, the bios may not be supplying enough voltage to the ram in dual channel mode, xmp is a generic setting and some ram has slightly substandard chips that require slightly more voltage than XMP will set. I've seen xmp set 1.35v, and the ram be unstable/non working, but manually raising to 1.37v allowed perfect stability.

And totally agree with the advice to wait. You Cannot diagnose full functionality of a component on a pc that isn't fully functional. That's like trying to guess how well a car will perform when it's missing a motor...

My old Ram was JEDEC and not xmp. I believe JEDEC is known to cause instability issues on some boards? The old Ram had a base voltage of 1.2v from memory. Most likely your theory about the bios not supplying enough voltage is correct, and probably could have been avoided by raising it. But then again, the Ram had a pretty bad latency of 22 and was only 3200mhz, so I really regret buying it in the first place. Very bad Ram kit for almost 400 bucks. It was the Team T create 64gb 2x 32. My board is also the Asrock z490 aqua, a gaming board, therefore the timings etc would have caused issues on creator style Ram I’m guessing. But I’m not expert.

 

[Karadjgne]

Jedec is a standard, basically from an organisation (Joint Electron Device Engineering Council) that sets standards and tables for ram use, speeds, timings, voltages etc. Most definitely will not cause instability. In other words your 3200 ram will have tables for 3200, 2933, 2666, 2400 etc. Jedec standards for DDR4 set base speeds of 2133MHz at 1.2v, and go up from there. Xmp is Intel eXtreme Memory Profile, an internal Intel setting for use in overclocking ram above what the cpu memory controller has been tested to. For DDR3, Intel tested the MC at 1333MHz/1600MHz, so any ram setting above that would be an XMP profile, but jedec tables would be provided for 1600MHz down to 800MHz.

As in all things, OC is no guarantee. It may have been tested on half a dozen different boards as working at the specified 3200MHz at stock xmp profile, at the factory, but that's no guarantee that with your cpu it might not work without manual manipulation of bios settings. Most likely you could have enabled xmp in the bios, then manually bumped one or more voltages and/or tightened up the timings.